Preprints of the 18th IFAC World Congress

Milano (Italy) August 28 - September 2, 2011

Extended State Observer for Nonlinear

Systems with Uncertainty
Bao-Zhu Guo Zhi-Liang Zhao

Academy of Mathematics and Systems Science, Academia Sinica,

Beijing 100190, P.R. China, and School of Computational and Applied
Mathematics, University of the Witwatersrand, Wits 2050,
Johannesburg, South Africa (e-mail: [email protected])

Department of Mathematics,University of Science and Technology of

China, Hefei 230061, P.R. China (e-mail: [email protected])

Abstract: The extended state observer first proposed by Jingqing Han in [J. Q. Han, The
extended state observer for a class of uncertain systems, Control and Decision, 10(1)(1995),
85-88] is the main interim loop of active disturbance rejection control that is taking off as a
technology after numerous successful applications in engineering. Unfortunately, there is no a
rigorous proof of convergence to date. In this paper, we attempt to tackle this long unsolved
extraordinary problem.
The main idea is to transform the error equation of objective system with its extended state
observer into a disturbed system of a asymptotical stable system, with which the effect of total
disturbance error is eliminated by high-gain.
Keywords: Extended state observer, nonlinear systems, uncertainty.
1. INTRODUCTION
The observer design is always one of fundamental issues
in nonlinear systems control. The Luenberger observerbased approach is the main method. They are many
other approaches like slide mode based approach. Some
of them are of advantage in robustness, but most of
them come up against the problems like adaptability
and chattering. We refer to Corless, Tu [1998], Darouach
et al. [1994], Gourshankar et al. [1997], Koshkouei, Zinober
[1998], Viswanadham, Ramakrishna [1980], Slotine et al.
[1987], Walcott et al. [1987], Walcott, Zak [1987] and
references therein, and the recent book Besancon [2007].
A particular attention is paid to Huang, Han [1999] where
the comparison of the approach discussed in present paper
with other existing approaches were presented in details.

x 1 (t) = x2 (t), x1 (0) = x10 ,

x 2 (t) = x3 (t), x2 (0) = x20 ,

.
..

x n (t) = f (t, x1 (t), x2 (t), . . . , xn (t))

+w(t) + u(t), xn (0) = xn0 ,

y(t) = x1 (t),

(2)

In his seminal work Han [1995], Han proposed the following extended state observer (ESO):

x
1 (t) = x
2 (t) 1 g1 (
x1 (t) y(t)),

x
2 (t) = x
3 (t) 2 g2 (
x1 (t) y(t)),

..
(1)
.

n (t) = x
n+1 (t) n gn (
x1 (t) y(t)) + u(t),

x
n+1 (t) = n+1 gn+1 (
x1 (t) y(t)),

where u is the input (control), y the output (measurement), f an usually unknown system function, and w
the uncertain external disturbance. f + w is called total disturbance in the elaborated expository paper Han
[2009]. (x10 , x20 , . . . , xn0 ) is the initial state, and i , i =
1, 2, . . . , n+1 are regulable constants. The main idea of the
extended state observer is that for the appropriately chosen functions gi , the state of the observer x
i , i = 1, 2, . . . , n
and x
n+1 can be, through regulating i , considered as
the approximations of the corresponding state xi for i =
1, 2, . . . , n, and the total disturbance f + w, respectively.
The last remarkable fact is the source in which the external
state observer is rooted. This idea is essentially the same
with the high-gain observer with augmented state variable
used in Freidovich, Khalil [2008] (an expository survey can
be found in Khalil [2008]). The numerical studies (e.g., Han
[1995]) and many other studies the years followed show
that for some nonlinear functions gi and parameter i ,
the observer (1) performs a very satisfactory adaptability,
robustness, and anti-chattering. For other perspectives of
this remarkable series researches, we refer the reader to
Han [2009].

which can be written as

Unfortunately, although huge applications have been carried out in engineering applications since then (see e.g.,
Hou et al. [2001], Miklosovic, Gao [2005], Zheng et al.
[2008], Zheng, Dong et al. [2007], Zheng, Gao [2010],

for an n-dimensional SISO nonlinear system

x(n) (t) = f (x(t), x(t),

. . . , x(n1) (t)) + w(t),

y(t) = x(t),

Copyright by the
International Federation of Automatic Control (IFAC)

1855

Preprints of the 18th IFAC World Congress

Milano (Italy) August 28 - September 2, 2011

Zheng, Cheng et al. [2007], Zhou, Gao [2007]) , the choice

of functions gi is essentially experiential. In order to apply
conveniently in practice, Gao [2003] proposed the following
linear extended state observer (LESO in short in what
follows), a special case of (1):

1
(y(t) x
1 (t)),
x
1 (t) = x
2 (t) +

1 (t)),
2 (t) = x
3 (t) + 2 (y(t) x
x

..
(3)
.

x
n (t) = x
n+1 (t) + n (y(t) x
1 (t)) + u(t),

n+1
x
n+1 (t) = n+1 (y(t) x
1 (t)),

where i , i = 1, 2, . . . , n + 1 are pertinent constants, and

is regulable gain constant.
The LESO (3) is essentially similar with the extended
high-gain observer studied in Freidovich, Khalil [2008].
High-gain observer has been extensively used to design
the stabilizing output feedback control by using singular perturbation method, see, for instance Atassi, Khalil
[2000], Esfandiari, Khalil [1992], Freidovich, Khalil [2008],
Oh, Khalil [1997] and references therein. The convergence
of the high-gain observer for determined systems without
augmented state variable was given in Theorem 6 of Besancon [2007] on page 18. The high-gain observer with
augmented state variable was used in Freidovich, Khalil
[2008] for tracking problems. A nice survey of high-gain
observer with or without augmented state variable can
be found in Khalil [2008]. However, the convergence of
high-gain observer with uncertainties, particularly with
augmented state variable is, to the best of our knowledge,
not available in literature. For linear extended state observer (3), Zheng et al. [2007] presented some convergence
results where either f + w is completely known or f + w
is completely unknown. The convergence results for some
special LESO were also available in Yang, Huang [2009].
In this paper, we are concerned with the convergence of
nonlinear extended state observe (NLESO in short in what
follows) of the following:

y(t) x
1 (t)

n1

,
1 (t) = x
2 (t) +
g1

y(t) x
1 (t)

n2
2 (t) = x

,
x

(t)
+

3
2

..
(4)
.

y(t) x
1 (t)

+ u(t),
n (t) = x
n+1 (t) + gn
x

1
yx
1 (t)

n+1 (t) = gn+1

,
x

n
which is a special case of (1) and an nonlinear generalization of LESO (3) for regulable gain parameter and
pertinent functions gi , i = 1, 2, . . . , n + 1. Notice that the
solution of (4) depends on parameter , but here we miss
by abuse of notation without confusion.
The following assumptions are presumed:

Assumption (H1). The possibly unknown functions f, w

are differentiable with respect to their variables, and

n
X
f f

c0 +

|u| + |f | + |w|
+ +
cj |xj |k
(5)
t
xi
j=1

for some positive constants cj and positive integer k.

We proceed as follows. In next section, Section 2, we

show that the NLESO (4) is convergent under additional
assumptions. Various special NLESO are presented in Section 3. Meanwhile, some Examples and numerical simulations are presented from time to time for illustrations.

2. THE CONVERGENCE OF EXTENDED STATE

OBSERVER
In this section, we consider the convergence of the NLESO
(4). To this end, we make some additional assumptions.
Assumption (H2). w and the solution of (2) satisfy |w|+
|xi (t)| B for some constant B > 0 and all i = 1, 2 . . . , n,
and t 0.

Assumption (H3). There exist constants i (i = 1, 2, 3, 4),

, and positive definite, continuous differentiable functions V, W : Rn+1 R such that
1 kyk2 V (y) 2 kyk2 , 3 kyk2 W (y) 4 kyk2 ,

n
X
V

(yi+1 gi (y1 ))

yi
i=1

yn+1 kyk,

V
gn+1 (y1 ) W (y),
yn+1

where y = (y1 , y2 , . . . , yn+1 ), k k denotes the Euclid norm

of Rn+1 .
Theorem 2.1. Suppose that Assumption (H1)-(H3) are
satisfied. Then
(i). For every positive constant a,
lim |xi (t) x
i (t)| = 0 uniformly in t [a, ).

(ii).

lim |xi (t) x

i (t)| O n+2i ,

where xi , x
i denote the solution of (2) and (4) respectively,
i = 1, 2, . . . , n + 1, xn+1 = f + w is the extended state
variable for system (2).
Proof. By extra state variable xn+1 = f +w (Han [2009]),
system (2) can be written as

x 1 (t) = x2 (t), x1 (0) = x10 ,

x 2 (t) = x3 (t), x2 (0) = x20 ,

..
.
(1)

x n (t) = xn+1 (t) + u(t), xn (0) = xn0 ,

x n+1 (t) = L(t), xn+1 (0) = L(0),

y(t) = x1 (t),
where L(t) = f (t, x1 (t), x2 (t), . . . , xn (t)) + w(t). We first
notice that

1856

Preprints of the 18th IFAC World Congress

Milano (Italy) August 28 - September 2, 2011

(t) =

f (t, x1 (t), . . . , xn (t))

t
n
X

f (t, x1 (t), . . . , xn (t))

xi+1 (t)
+
x
i
i=1
=

identity matrix I. Define the Lyapunov function V, W :

Rn+1 R by

d
f (s, x1 (s), . . . , xn (s))
+ w(t)

ds
s=t

V () = hP , i,
2

(2)

Set

ei (t)
, i = 1, 2, . . . , n + 1.
n+1i
(3)
Then a direct computation shows that = (1 , 2 , . . . ,
n+1 ) satisfies

e1 (0)

,
1 = 2 g1 (1 ), 1 (0) =

e2 (0)

2 = 3 g2 (1 ), 2 (0) = r1 ,

(4)
..

en (0)

,
n = n+1 gn (1 ), n (0) =

n+1 = gn+1 (1 ) + (t), n+1 (0) = er+1 (0).

ei (t) = xi (t) x
i (t), i (t) =

By Assumption (H3), finding the derivative of V ((t)) with

respect to t along the solution (t) of system (4) gives

p
3
1
d
(5)
M V ().
V ((t)) V () +
dt
2
1
It follows that

dp
1 M
3 p
V ((t))
V ((t)) +
.
(6)
dt
22
21
By Assumption (H3) again, we have
p
Z
3
3
1 V ((0)) 2
M t 2
t
(ts)
2 +
1
e
e
ds. (7)
kk
1
21 0
This together with (3) yields
"p
3 t
1 V ((0)) 2
n+1i
2
|ei (t)|
e
1
(8)
#
Z t
3
M 2
(t/s)
1
+
e
ds .
21 0

Both (i) and (ii) of Theorem 2.1 then follow from (8). The
proof is complete.
2
The result of Theorem 2.1 enables us to deduce the
convergence of LESO (3) immediately. Actually, if the
matrix following is Hurwitz:

1 1 0 0
2 0 1 0
.
.. .. . . ..

E=
(9)
. . . . ,
..
0 0 1
n
n+1 0 0 0
let P be the positive definite matrix solution of the Lyapunov equation P E + E P = I for (n + 1)-dimensional

(10)

Then min (P )kk V () max (P )kk ,

n
X
V
i=1

+u(t)
f (t, x1 (t), . . . , xn (t)) + w(t).

xn
From Assumptions (H1)-(H2), |(t)| M is uniformly
bounded for some M > 0 and all t 0.

W () = h, i.

(i+1 i 1 )

V
n+1 1 = kk2 = W (),
n+1

and

V V

n+1 = k2 P k 2kP kkk = 2max (P )kk.

where max (P ) and min (P ) are the maximal and minimal

eigenvalues of P , respectively. So V, W defined by (10)
satisfies Assumption (H3). Therefore, as a consequence of
Theorem 2.1, we have improved the stability of LESO (3)
presented in Zheng et al. [2007] where the a in the following
Corollary 2.1 is required to be large.
Corollary 2.1. If E is Hurwitz and assumptions (H1)-(H2)
are satisfied, then
(i). For every positive constant a,
lim |xi (t) x
i (t)| = 0 uniformly in t [a, ).

(ii).

lim |xi (t) x

i (t)| O n+2i ,

where xi , x
i denote the solutions of (2), (3) respectively,
i = 1, 2 , n + 1, xn+1 = f + w is the extended state
variable for system (2).
Example 2.1. For the system following

x (t) = x2 (t),

1
x 2 (t) = x1 (t) x2 (t) + w(t) + u(t),
(11)

y(t) = x1 (t),

we design an nonlinear extended state observer following:

3
y(t) x
1 (t)

x
1 (t) = x
2 (t) + (y(t) x
1 (t))
,

3
2 (t) = x
3 (t) + 2 (y(t) x
1 (t)) + u(t),
x

1
x
1 (t)).
3 (t) = 3 (y(t) x

(12)
1
where : R R is defined as (r) = for r

1
,
, (r) = sin r for r ,
, (r) =
2
4
2 2

for r
, . In this case, gi in (4) can be taken
4
2
g1 (y1 ) = 3y1 (y1 ), g2 (y1 ) = 3y1 , g3 (y1 ) = y1 . (13)

The Lyapunov function V : R3 R for this case is defined

by
Z x1
(s)ds,
(14)
V (y) = hP y, yi +
0

where P is the positive definite solution

3 1
equation P E + E P = I with E = 3 0
1 0
computation shows that

1857

of!Lyapunov
0
1 . A direct
0

Preprints of the 18th IFAC World Congress

Milano (Italy) August 28 - September 2, 2011

2
X
V
i=1

V
(yi+1 + gi (y1 )) +
g3 (y1 )
yi
y3
2

y1
7y 2
3y 2

+ 2 + 3 , W (y1 , y2 , y3 ).
8
8
4

v(t) x
1 (t)

n1

1 (t) = x
,
x

(t)
+

2
1

v(t) x
1 (t)

n2

,
x
2 (t) = x
3 (t) +
g2

..
.

v(t) x
1 (t)

n (t) = x

(t)
+
g
,

n+1
n

v(t) x
1 (t)
1

,
g
x

(t)
=
n+1
n+1

(15)

So all conditions of Assumption (H3) are satisfied. Therefore, (12) serves as a well defined NLESO for (11) according to Theorem 2.1. Now take the data as
x1 (0) = x2 (0) = 1, x
1 (0) = x
2 (0) = x
3 (0) = 0,
u(t) = sin t, w(t) = cos t, = 0.01.

(16)

The simulations for NLESO (12) are plotted as Fig. 1.

It is seen from Fig. 1 that the NLESO (12) is well tracking
the state of the system (11).
In what follows, we relax the conditions of Assumption
(H3) by Assumption (H4) following.
Assumption (H4). There exist constant R > 0, > 0
and positive definite, continuous differentiable function
V, W : Rn+1 R such that
{y|V (y) d} is bounded for any d > 0,

V
V
(yi+1 gi (y1 ))
gn+1 (y1 ) W (y),
y
y
i
n+1
i=1

yn+1 W (y) for kyk > R.

|xi (t) x
i (t)| < , t (T , ),
where T > 0 depends on , xi , x
i denote the solutions of
(2), (4) respectively, i = 1, 2 , n + 1, xn+1 = f + w is
the extended state variable for system (2).
2

3. SOME SPECIAL EXTENDED STATE OBSERVER

In this section, we state some applications of extended
state observer. The first special case is what so called output tracking differentiator, a special kind of tracking differentiator (Han [1989, 2009]). Suppose that v is the tracked
signal. Let xi (t) = v (i1) (t). Then xi , i = 1, 2, . . . , n satisfy

x 1 (t) = x2 (t),

x 2 (t) = x3 (t),

..
(17)
.

(n1)

(t),

x n (t) = v
y(t) = x1 (t) = v(t).
The corresponding NLESO (4) becomes

(i). For every positive constant a,

lim |v (i1) (t) x
i (t)| = 0,

uniformly on [a, ).

lim |v (i1) (t) x

i (t)| O n+2i ,

where x
i is the solution of (18), i = 1, 2 , n + 1.

We can obtain the weak convergence result stated in

Theorem 2.2 below.
Theorem 2.2. Under Assumptions (H1),(H2) and (H4),
the nonlinear extended state observer (4) is convergent
in the sense that for any (0, 1), there exists (0, 1)
such that for any (0, ),

Duo to page limiitation, we omit the proof.

The convergence result in this special case is stated as

Theorem 3.1 below.
Theorem 3.1. [Output tracking differentiator] Suppose that Assumption (H3) holds and v (n) (t) is bounded.
Then

(ii).

n
X

(18)

The next example is different to system (2) and those

discussed in Zhou et al. [2009], in which the system
function f is known. The unknown part is only the external
disturbance w. In this case, our NLESO can be modified
as

x1 (t) x
1 (t)
n1

1 (t) = x
2 (t) +
g1
,
x

x1 (t) x
1 (t)

n2

2 (t) = x

,
x

(t)
+

g
3
2

..
.
(19)

x
(t)

(t)

1
1

x
n (t) = x
n+1 (t) + gn

+f (t, x
1 (t), x
2 (t), . . . , x
n (t)) + u(t),

x1 (t) x
1 (t)

x
.
n+1 (t) = gn+1
n
Using the same notation as in (3) and setting xn+1 = w
in this case, we get

e (0)

1 = 2 g1 (1 ), 1 (0) = 1 ,

e2 (0)

2 = 3 g2 (1 ), 2 (0) = n1 ,

..

en (0)

n = n+1 gn (1 ) + 1 (t), n (0) = ,

n+1 = gn+1 (1 ) + 2 (t), n+1 (0) = en+1 (0),

(20)
where

1858

Preprints of the 18th IFAC World Congress

Milano (Italy) August 28 - September 2, 2011

1 (t) = f (t, x1 (t), . . . , xn (t)) f (t, x

1 (t), . . . , x
n (t)),
2 (t) = w(t).

(21)
Theorem 3.2. [Modified extended state observer] In
addition
the conditions in Assumption (H3), we assume
to
V
that y
kyk, L < 3 , where L is the Lipschitz
n
constant of f :
|f (t, x1 , . . . , xn ) f (t, y1 , . . . , yn )| Lkx yk,
(22)
t 0, x = (x1 , x2 , . . . , xn ) , y = (y1 , y2 , . . . , yn ) Rn .

If w is bounded in (2), then

(i). For every positive constant a,

lim |xi (t) x
i (t)| = 0 uniformly in t [a, )
0

(ii).

lim |xi (t) x

i (t)| O n+2i ,

where xi , x
i denote the solutions of (2), (19) respectively,
i = 1, 2, . . . , n + 1, xn+1 = w.
Proof. Finding the derivative of V ((t)) with respect to t
along the solution (t) of system (20) yields

p
1
3 L
d
V ((t))
V () +
M V (). (23)
dt
2
1
It follows that

3 L p
dp
1 M
V ((t))
V ((t)) +
. (24)
dt
22
21
This together with Assumption (H3) gives
p
Z
L
L
1 V ((0)) 32
M t 32
t
(ts)
2
1
+
ds.
e
e
kk
1
21 0
(25)
By (3), it follows that
"p
1 V ((0)) (32L)t
2
|ei (t)| n+1i
e
1
(26)
#
Z t
L
M 32
(t/s)
1
ds .
+
e
21 0

The (i) and (ii) of Theorem 3.2 follow from (26). The proof
is complete.
2
Example 3.1. For the system following

x 1 (t) = x2 (t),
(27)
x 2 (t) = sin(x1 (t)) + sin(x2 (t)) + w(t) + u(t),
4
where w is the external disturbance, we design the corresponding modified linear extended state observer as:

1 (t)),
x
1 (t) = x
2 (t) + (y(t) x

11

2 (t) = x
3 (t) + 2 (y(t) x
1 (t))
x

(28)

sin(
x1 (t)) + sin(
x2 (t))

+
+ u(t),

x
3 (t) = 3 (y(t) x
1 (t)).

!
6 1 0
For this example, the associated matrix E = 11 0 1
6 0 0
has three eigenvalues {1, 2, 3}, so it is Hurwitz. Use

Matlab to solve the Lyapunov equation P E + E P =

I, to find that the eigenvalues of P satisfy max (P )
2.3230 < . Let V, W : Rn+1 R be defined as
2
V () = hP , i, W () = h,
i.

Then min (P )kk

V
kV ()k max (P )kk2 ,
2max (P )kk, and

V
V
V
(2 61 ) +
(3 111 ) 61
= W (). Now
1
2
1
1
2 (t))
f (x1 , x2 ) = sin(x1 (t))+sin(x
, and we find that L = 2
.
4
1
Hence Lmax (P ) < 2 . Therefore, for any bounded control
u and bounded disturbance w, w (lets say the
Pp finite superposition of sinusoidal disturbance w(t) = i=1 ai sin bi t),
by Theorem 3.2, for any a > 0 lim0 |xi (t) x
i (t)| =
i (t)|
0 uniformly for t [a, ) and lim |xi (t) x
t

O n+2i . Once again, we use the same data presented

in (16) to simulate Example 3.1. The results are plotted
in Fig. 2. It is seen from Fig.2 that the modified extended
state observer (28) tracks very satisfactorily the state of
system (27).
The last case is more special that the system function f
is known and the external w = 0. In this case, system
(2) becomes a special deterministic nonlinear system. To
design the observer, we make the following assumption.
Assumption (H5). f is locally Lipschitz continuous, and
n
X

f (t, x)

h(t, x) = f (t, x) +
f (t, x) +
u(t).
xi+1 (t)
t
x
xn
i
i=1
(29)
is globally Lipschitz continuous in x = (x1 , x2 , . . . , xn )
uniformly for t (0, ), where xn+1 (t) = f (t, x(t)).

The state observer for (2) now is designed as following:

y(t) x
1 (t)

n1

,
x
1 (t) = x
2 (t) +
g1

y(t) x
1 (t)

n2
2 (t) = x

,
x

(t)
+

3
2

..
(30)
.

y(t) x
1 (t)

+ u(t),
n (t) = x
n+1 (t) + gn

x
n

yx
1 (t)
1

x
.
n+1 (t) = h(t, x
) + gn+1

n
Theorem 3.3. Under assumptions (H3) and (H5), it has
(i). There exists an 0 > 0 such that for any (0, 0 )
lim |xi (t) x
i (t)| = 0,
t

(ii). For any a > 0,

lim |xi (t) x
i (t)| = 0 uniformly in t [a, )
0

where xi , x
i are the solutions of (2), (30) respectively,
i = 1, 2, . . . , n + 1, xn+1 (t) = f (t, x(t)).
Duo to page limitation, we omit the proof.

4. CONCLUDING REMARK
In this paper, the convergence of various high gain nonlinear extended state observers for a kind of nonlinear

1859

Preprints of the 18th IFAC World Congress

Milano (Italy) August 28 - September 2, 2011

systems with uncertainty both from internal and external

is proved. The high gain is used to eliminate the influence
of uncertainties. Examples and numerical analysis are followed to illustrate the effectiveness of the corresponding
observers. Due to high gain and the difference between the
initial states of the tracked system and the observer, the
peaking phenomena observed in Freidovich, Khalil [2008]
in the beginning of the time interval is also observed from
the numerical analysis.
REFERENCES
A. N. Atass and H.K. Khalil. Separation results for the
stabilization of nonlinear systems using different highgain observer designs. Sys. & Contr. Lett., 39:183191,
2000.
G. Besancon. Nonlinear Observers and Applications.
Springer Verlag, New York, 2007.
M.J. Corless and J. Tu. State/input estimation for a class
of uncertain systems. Automatica, 34:757764, 1998.
M. Darouach, M. Zasadzinski, and J.S. Xu. Full-order
observers for linear systems with uncertain inputs. IEEE
Trans. Auto. Contr., 39:606609, 1994.
F. Esfandiari and H.K. Khalil. Output feedback stabilization of fully linearizable systems. Int. J. Contr., 56:
10071037, 1992.
L. B. Freidovich and H. K. Khalil. Performance recovery
of feedback-linearization-based Designs. IEEE. Trans.
Auto. Contr., 53:23242334, 2008.
Z. Gao. Scaling and bandwith-parameterization based
controller tuning. American Control Conference, 4989
4996, 2003.
V. Gourshankar, P. Kudva, and K. Ramar. Reduced order observer for multivariable systems with inaccessible
disturbance inputs. Int. J. Contr., 25:311-319, 1977.
J. Q. Han. Control theory: Model approach or control
approach. J. Sys. Sci. & Math. Scis., 9(4):328335,
1989(in Chinese).
J. Q. Han. A class of extended state observers for uncertain
systems. Control & Decision, 10(1):85-88, 1995(in
Chinese).
J. Q. Han. From PID to active disturbance rejection
control. IEEE Trans. Ind. Electron., 56:900-906, 2009.
Y. Hou, Z. Gao, F. Jiang, and B.T. Boulter. Active
disturbance rejection control for web tension regulation.
IEEE Conference on Decision and Control , 4974-4979,
2001.
Y. Huang and J. Q. Han. A new synthesis method for
uncertain systems-the self-stable region approach. International Journal of Systems Science, 30:33-38, 1999.
H. K. Khalil. High-gain observers in nonlinear feedback
control. Int. Conf. Control, Automation and Systems,
Oct.14-17, Korea, 2008.
A. Koshkouei and A. Zinober. Sliding mode controller
observer design for SISO linear systems. Int. J. Sys.
Sci., 29:1363-1373, 1998.
P. Kudva, N. Viswanadham, and A. Ramakrishna. Observers for linear systems with unknown inputs. IEEE
Trans. Auto. Contr., 25:113-115, 1980.
R. Miklosovic and Z. Gao.
A dynamic decoupling
method for controlling high performance turbofan engines. Proc. of the 16th IFAC World Congress, July
4-8, 2005, Czech Republic.

S. Oh and H.K. Khalil. Nonlinear output-feedback tracking using high-gain observer and variable structure control. Automatica, 33:1845-1856, 1997.
J. Slotine, J. Hedrick, and E. Misawa. On sliding observers
for nonlinear systems. ASME J. Dynam. Sys. Measur.
& Contr., 109:245-252, 1987.
B.L. Walcott, M.J. Corless, and S.H. Zak. Comparative
study of non-linear state-observation technique. Int. J.
Contr., 45:2109-2132, 1987.
B. L. Walcott and S.H. Zak. State observation of nonlinear
uncertain dynamical systems.
IEEE Trans. Auto.
Contr., 32:166-170, 1987.
X. X. Yang and Y. Huang. Capability of extended state
observer for estimating uncertainties. American Control
Conferencce, 3700-3705, 2009.
Q. Zheng, L.Gao, and Z. Gao. On stability analysis
of active disturbance rejection control for nonlinear
time-varying plants with unknow dynamics.
IEEE
Conference on Decision and Control, 3501-3506, 2007.
Q. Zheng, L. Gong, D.H. Lee, and Z. Gao. Active
disturbance rejection control for MEMS gyroscopes.
American Control Conference, 4425-4430, 2008.
Q. Zheng, L.L. Dong, and Z. Gao. Control and rotation rate estimation of vibrational MEMS gyroscopes.
IEEE Multi-conference on Systems and Control, 118123, 2007.
Q. Zheng and Z.Gao. On applications of active disturbance
rejection control. Chinese Control Conference, 60956100, 2010.
Q. Zheng, Z. Chen, and Z. Gao. A dynamic decoupling
control approach and tts applications to chemical processes. American Control Conference, 5176-5181, 2007.
W. Zhou and Z. Gao. An active disturbance rejection
approach to tension and velocity regulations in Web
processing lines. IEEE Multi-conference on Systems
and Control, 842-848, 2007.
W. Zhou, S. Shao, and Z. Gao. A stability study of
the active disturbance rejection control problem by a
singular perturbation approach. Appl. Math. Sci., 3:
491-508, 2009.
2

1.5

1
1

0.5
0

0.5

1.5

4
0

10

4
0

(a)
x1 (green), x
1 ( red),
1 (blue)
x1 x

10

(b)
x2 (green), x
2 (red),
2 (blue)
x2 x

10

(c)
x3 (green), x
3 ( red),
3 (blue)
x3 x

Fig. 1. NLESO (12) for system (11)

30

4
3

25

3
2

20
2
1
15
1

10
1
0
5
2
1

2
0

(a)
x1 (green), x
1 (red),
x1 x
1 (blue)

10

4
0

(b)
x2 (green), x
2 (red),
x2 x
2 (blue)

10

10

(c)
x3 (green), x
3 (red),
x3 x
3 (blue)

Fig. 2. The modified extended state observer (28) for

system (27)

1860